MadewellRD
UserVendor-agnostic agent skill suites for the software lifecycle, web, AI engineering, product, sales, and mobile. Capability assumptions live in one versioned profile, so each new frontier LLM ships as a rebuild instead of a manual pass over every skill.
Categories
Indexed Skills (50)
agent-architecture-desk
design AI agent architecture including planning boundaries, execution loops, memory and state strategy, tool routing, approval gates, retries, delegation, and halt behavior.
agent-observability-desk
design observability for AI agents and workflows including traces, prompts, model calls, tool calls, retrieval events, approvals, errors, eval probes, cost, latency, and safety signals.
ai-engineering-command-desk
orchestrate AI engineering workflows from capability intent through model, prompt, tool, agent, retrieval, eval, safety, inference, observability, release, and incident stages using connector-grounded evidence, workflow packets, stage advancement, and halt behavior.
ai-incident-response-desk
triage AI production incidents involving hallucination spikes, safety failures, prompt injection, tool misuse, data leakage, model regressions, cost spikes, latency degradation, eval regressions, or user harm reports.
ai-release-readiness-desk
assess readiness to release AI capabilities across requirements, evals, safety review, red-team status, inference ops, observability, rollback, docs, support handoff, and owner approval.
ai-safety-review-desk
review AI capability risks including misuse, policy compliance, privacy, security, hallucination harm, data leakage, autonomy, tool-use risk, user impact, and mitigations.
cost-latency-optimization-desk
optimize AI system cost and latency using model routing, caching, prompt compression, context pruning, batching, streaming, parallelism, retrieval tuning, and fallback tiers while preserving quality and safety gates.
dataset-curation-desk
plan and review AI datasets for source selection, labeling, balancing, privacy, deduplication, train and eval splits, drift, provenance, consent, and retention.
eval-design-desk
design AI evaluation plans with goals, datasets, rubrics, grading methods, thresholds, regression slices, safety checks, human review, and reporting requirements.
eval-run-analysis-desk
analyze completed AI eval runs, regression deltas, failure clusters, grading reliability, threshold status, release blockers, and rerun recommendations.
fine-tuning-desk
assess and plan fine tuning only when prompt, retrieval, tool, model routing, and eval evidence justify training a specialized model.
inference-ops-desk
plan production inference operations including deployment topology, rate limits, quotas, retries, caching, streaming, fallbacks, batching, timeouts, secrets, logging, and SLOs.
model-selection-desk
select model candidates, routing constraints, fallback behavior, and model tradeoffs for AI capabilities using task fit, quality, latency, cost, safety, modality, context, and deployment evidence.
prompt-systems-desk
design prompt systems, instruction hierarchy, context assembly, prompt contracts, refusal and defer behavior, prompt evaluation fixtures, prompt injection defenses, and prompt observability hooks for AI capabilities.
red-team-eval-desk
plan and analyze adversarial AI testing for jailbreaks, prompt injection, data exfiltration, harmful instructions, over-permissioned tools, and policy evasion.
retrieval-rag-design-desk
design retrieval augmented generation systems with indexing, chunking, embeddings, ranking, filters, citations, freshness policy, permission filtering, and grounding behavior.
synthetic-data-desk
design synthetic data generation workflows with seed examples, constraints, diversity targets, contamination controls, review loops, and validation gates.
tool-schema-design-desk
design AI tool schemas, resource contracts, permission boundaries, argument validation, idempotency rules, error semantics, and result contracts for agentic workflows.
android-app-engineering-desk
prepare Android native app implementation plans for Kotlin, Java, Jetpack Compose, View systems, modularization, storage, networking, background work, sensors, permissions, and platform APIs.
android-architecture-design-desk
design Android app and game architecture, module boundaries, data flow, offline behavior, engine integration, services, APIs, migrations, and ADR-ready decisions.
android-backend-integration-desk
define Android service and API integration, auth, sync, payments, push notifications, analytics, remote config, multiplayer, leaderboards, cloud saves, retries, offline behavior, and failure modes.
android-command-desk
orchestrate complete Android app and game development workflows across discovery, product, architecture, implementation, testing, release, Play Store operations, live ops, and maintenance. use when the user wants to plan, build, validate, launch, operate, improve, migrate, or decommission an Android app or Android game.
android-game-engineering-desk
prepare Android game implementation plans for AGDK, NDK, C/C++, Unity, Unreal, Godot, custom engines, rendering, input, assets, frame pacing, and gameplay/runtime constraints.
android-maintenance-growth-desk
plan Android maintenance, dependency upgrades, SDK target updates, deprecations, Play policy changes, experiments, monetization iteration, store optimization, retention, and technical debt.
android-observability-liveops-desk
define Android observability and live ops for crash reporting, logs, metrics, analytics events, alerts, feature flags, remote config, game economy/events, rollout monitoring, and incident response.
android-performance-optimization-desk
plan Android performance for startup, memory, battery, ANR and crash risk, rendering, frame pacing, asset loading, Macrobenchmark, Baseline Profiles, profiling, and device-tier budgets.
android-product-requirements-desk
define Android app and game product requirements, audience, platform targets, acceptance criteria, non-goals, risks, Play constraints, monetization assumptions, and open questions.
android-release-store-ops-desk
plan Android builds, signing, versioning, CI/CD, AAB/APK packaging, internal testing, Play tracks, release notes, staged rollout, rollback, Play Asset Delivery, and store listing readiness.
android-security-privacy-desk
review Android security, privacy, permissions, secrets, Play policy risk, data safety, secure storage, anti-tamper, networking, dependency risk, and abuse controls.
android-technical-discovery-desk
inspect Android repo, Gradle, SDK, NDK, dependency, manifest, device, emulator, engine, CI, feasibility, constraint, and unknown facts before implementation.
android-testing-qa-desk
define Android app and game QA, unit tests, instrumented tests, UI tests, screenshot tests, device matrix, emulator and physical coverage, gameplay smoke, regression, and release gates.
android-ui-ux-desk
plan Android UI/UX, Material design, navigation, responsive layouts, accessibility, input modes, localization, onboarding, and app or game interaction states.
cloud-cost-rightsizing-desk
build the cloud cost allocation model and allocable share, set budgets and anomaly thresholds with named recipients, identify rightsizing candidates from percentile utilization evidence with the performance risk each carries, measure commitment and reservation coverage and utilization against a stable baseline, plan storage tiering and retention savings net of retrieval and early-delete cost, reclaim idle and orphaned resources behind dependency checks, and define unit cost metrics the business recognizes.
cloud-decommissioning-desk
retire cloud resources stacks accounts and regions safely using an evidence-backed dependent inventory from flow and access and authentication logs, a notice window with named owners, a reversible quarantine step before deletion, data disposition against retention and legal holds, ordered teardown with the irreversible boundary marked, credential revocation and address and dns release in the order that prevents takeover, removal of the code and pipeline entries that would recreate it, and confirmation that the billing line actually stopped.
cloud-identity-access-desk
design cloud identity and access, covering federation and single sign-on for human access, role and permission-set structure with least privilege, permission boundaries and their interaction with organization-level denies, workload identity that removes static access keys, cross-account trust and its direction, standing and privileged access findings, break-glass with storage and alerting, and access review cadence with named reviewers. use for cloud iam design, federation rollout, least-privilege reduction, trust policy review, static credential elimination, and access recertification.
cloud-infrastructure-command-desk
orchestrate cloud infrastructure work across landing zones, account and subscription structure, cloud iam and federation, vpc and cidr network topology, hybrid connectivity and dns, compute and managed kubernetes platforms, object storage and managed databases, multi-region resilience and disaster recovery, infrastructure as code and state backends, provisioning pipelines and plan approval, key management and secret rotation, cloud security posture and cis benchmarks, tagging and resource inventory, cost allocation rightsizing and savings commitments, drift detection, migration waves, and decommissioning. use when the user wants to design, provision, harden, reconcile, rightsize, migrate, or retire cloud infrastructure in one or more providers.
cloud-migration-desk
plan cloud migration from source estate discovery and the dependency graph through disposition per workload across rehost replatform refactor repurchase retain and retire, wave sequencing derived from coupling, landing readiness and target quota per wave, data migration method with lag-driven cutover windows, the rollback boundary past which rollback stops existing, coexistence and dual-running behavior, and post-cutover validation against a baseline captured before the move.
cloud-network-architecture-desk
design cloud network architecture, covering the address allocation register and cidr planning, virtual network and subnet layout across availability zones, hub-and-spoke or transit topology and its routing consequences, segmentation with security group and firewall policy structure, the egress model and centralized inspection, private service endpoints, load balancer and ingress tiers, and the reachability matrix stating which segments may reach which. use for vpc and vnet design, address planning, subnet sizing, transit routing, microsegmentation, egress inspection, and private endpoint placement.
cloud-security-posture-desk
assess cloud security posture against named benchmark controls with reachable-exposure analysis rather than raw finding counts, covering public exposure across storage and compute and database and network surfaces, encryption and audit logging coverage per account and region, guardrail coverage gaps where a control exists in policy but at no enforcement point, finding prioritization by exposure path, the exception register with named owners and expiry dates, and a remediation plan mapped to change class.
cloud-storage-data-services-desk
design cloud storage and data services, covering object block and file storage selection per access pattern, bucket share and volume access policy, lifecycle and tiering rules with their retrieval cost and time, versioning object lock and immutability for retention obligations, public-access blocking and its enforcement point, encryption with named key ownership, backup destination frequency and retention including isolated copies, and the restore path with the date it was last exercised. use for storage selection, bucket policy review, lifecycle and archive tiering, worm and legal hold, backup design, and restore testing.
cloud-workload-intake-desk
frame a cloud workload before any infrastructure is designed, covering criticality tiering, data classification, residency and sovereignty constraints, the compliance regimes actually in scope, rto and rpo objectives and whether they are commitments or aspirations, budget envelope, provider and region candidacy, managed-service versus self-operated disposition, and explicit non-goals. use at the start of a new workload, a migration intake, a landing zone request, or any estate change whose requirements were never written down.
compute-platform-desk
select and size cloud compute, covering platform choice across virtual machines autoscaling groups serverless functions and managed container runtimes, instance family and size rationale from utilization evidence, machine image lineage and the rebuild path, autoscaling triggers bounds and cooldowns with health check grace, interruptible capacity mix and its resilience cost, placement across availability zones and failure domains, and the patch and provider-forced upgrade path. use for compute platform selection, instance sizing, autoscaling policy, golden image pipelines, spot capacity strategy, and instance retirement planning.
configuration-secrets-desk
design cloud key management and secrets handling including the key hierarchy with scope and ownership and rotation state, envelope encryption and key policy, secret store selection and per-consumer access policy, rotation with dual-slot cutover, short-lived workload credentials replacing static access keys, configuration layering and precedence across environments, secret delivery into compute and container and pipeline surfaces, and remediation of credentials found in state files, machine images, or logs.
container-platform-desk
design the managed kubernetes and container platform, covering cluster topology and how many clusters exist for what reason, control plane and node group configuration, node autoscaling and bin packing with requests limits and disruption budgets, cluster version upgrades against the provider support window and removed apis, ingress and service exposure, container registry and image provenance, admission control and in-cluster isolation, storage classes, and the platform-versus-workload ownership boundary. use for cluster design, node pool sizing, cluster upgrade planning, ingress architecture, registry and image signing, and multi-tenancy inside a cluster.
drift-detection-reconciliation-desk
detect and reconcile cloud infrastructure drift by comparing declared state in code against live provider state on a defined cadence, score the drift inventory by consequence rather than count, attribute each change to its actual origin from audit and activity log evidence, suppress provider-side diff noise deliberately, decide reconciliation disposition across codify and revert and adopt and accept, import unmanaged resources safely, and change the guardrail that stops the drift recurring.
hybrid-connectivity-dns-desk
design hybrid connectivity and dns, covering dedicated circuits and encrypted tunnels with genuinely diverse redundant paths, bgp route exchange and failover behavior including degraded-bandwidth backup, on-premises address overlap resolution, dns zone architecture across the hybrid boundary with split-horizon resolution forwarders and resolver endpoints, certificate ownership and renewal, and global traffic distribution with health checks and ttl behavior. use for direct circuit design, vpn failover, bgp routing, hybrid dns, split-horizon zones, certificate lifecycle, and latency or geo traffic steering.
infrastructure-as-code-desk
design infrastructure as code repository and stack layout including state boundaries and blast radius, module interfaces and semantic versioning, remote state backend location with locking and encryption and read access, provider and dependency version pinning with lock files, composition patterns across environments, validation and policy unit test gates, secrets-out-of-state discipline, state import and refactor safety, and measured codification coverage of the live estate.
landing-zone-account-structure-desk
design the cloud organization hierarchy and account structure, covering organizational units folders and management groups, account subscription and project separation by environment and sensitivity, account vending and the day-one baseline, organization-level deny policies and their attachment points, region enablement and restriction, centralized log archive security and network accounts, and the audit logging config recording backup and threat detection every account carries. use for landing zone design, new account or subscription requests, organizational unit restructuring, and guardrail attachment review.
managed-database-platform-desk
design managed database platforms, covering engine and service selection against the access pattern, instance sizing and storage configuration, high-availability topology across failure domains, read replica placement and replication lag, backup and point-in-time recovery windows measured against stated rto and rpo, parameter baselines and the static settings that require a reboot, connection limits and pooling, major version upgrades against provider end-of-support dates, and database credential handling. use for database engine selection, ha topology, replica design, pitr and backup windows, parameter tuning, connection exhaustion, and version end-of-life upgrades.
provisioning-pipeline-desk
design the infrastructure provisioning pipeline including plan generation and the plan review gate, policy-as-code evaluation against the machine-readable plan, the approval matrix keyed to blast radius, a least-privileged federated apply identity per environment, the rule that the reviewed plan artifact is the applied artifact, environment promotion and permitted divergence, concurrency and state lock behavior, the sanctioned manual change path, and the rollback boundary per stack.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.